PubAg

Main content area

Testate amoebae as a hydrological proxy for reconstructing water-table depth in the mires of south-eastern Australia

Author:
Zheng, Xianglin, Amesbury, Matthew J., Hope, Geoffrey, Martin, Len F., Mooney, Scott D.
Source:
Ecological indicators 2019 v.96 pp. 701-710
ISSN:
1470-160X
Subject:
altitude, autocorrelation, ecology, environmental indicators, peatlands, prediction, water table, Australia
Abstract:
Although it is well established that moisture availability in south-eastern Australia has been decreasing through time recently, the driver(s) of this trend are contentious, and our understanding of any drivers is limited by a relatively short historic record. Testate amoebae have been widely used to reconstruct peatland hydrology in the Northern Hemisphere, but in the Southern Hemisphere research is still needed to assess their proficiency as a palaeohydrological proxy and to develop robust transfer functions. Here we examine the ecology of testate amoebae in several high altitude mires in south-eastern Australia and present the first transfer function for the continent. Euglypha tuberculata type, Centropyxis platystoma type and Assulina muscorum were the most common taxa in our modern samples. Water-table depth was the primary environmental variable determining testate amoebae assemblages and therefore transfer functions were developed for this ecological factor. We found that the performance of various all-species and species-pruned transfer functions were statistically robust, with R2 values of around 0.8 and Root Mean Squared Error of Prediction (RMSEP) values of about 7 cm. All cross-validation methods (leave-one-out RMSEP, cluster-bootstrap RMSEP, segment-wise RMSEP and leave-one-site-out RMSEP from all-species and species-pruned transfer functions) suggested that the Modern Analogue Technique (MAT) was the best performing transfer function, with negligible bias evident from un-even sampling and spatial autocorrelation. We also used a new approach to evaluate the importance of taxa and the performance of our transfer functions using species-pruned methods. Our results suggest that the all-species MAT, with an RMSEP of 5.73 and R2 of 0.86, provides the best reconstruction of water-table depth across our sites in south-eastern Australia.
Agid:
6252556